Governance, transparency, e governance and the civil services

Eligibility is now settled by data matching, not by an official's visit. Aadhaar seeding removed crores of duplicate names. The saving is real, and so is the silence when the system refuses.

The accountability questions

One rule, lakhs excluded: A human error affects one file. A faulty rule reaches every counter at once. Deaths were reported in Jharkhand after ration was denied for want of Aadhaar linkage.

No reasons, so no appeal: An applicant told only that the system rejected him cannot frame a challenge. Natural justice requires notice and reasons.

The official defers to the screen: A verifier trusts the database over the person before him. Discretion survives on paper and disappears in practice.

Confident answers from stale data: The Socio Economic and Caste Census was conducted in 2011. A household ranked on it may have changed since.

Two warnings from abroad: A Dutch court struck down the SyRI welfare risk system in 2020. Australia's Robodebt inquiry found debts raised unlawfully.

Safeguards

A person signs every refusal: The model may flag a case. Only an officer should record the decision, with reasons.

Measure exclusion, not only fraud: Departments count savings and ghost names removed. Publishing false exclusions by district would show the other side.

The tests already exist: Puttaswamy, 2017 requires a legitimate aim and the least intrusive means. The DPDP Act, 2023 limits data to its stated purpose.

An algorithm can find a duplicate. It cannot hear an explanation. Where a benefit is withdrawn, the machine may advise and a person must decide.

Decode

Two demands. Examine the accountability questions, then suggest safeguards.

Implicit demand

Algorithmic error differs in kind. It is silent, and it repeats at scale.

Architecture

What it does → where it is used → the accountability questions → safeguards → verdict.

Articulation

Anchor: Aadhaar deduplication, SyRI, Robodebt, the DPDP Act, Puttaswamy. Money line: the machine may advise and a person must decide.

Where it is used

Automated data matching has been used by states to identify ineligible ration card holders.

Socio Economic and Caste Census data has driven targeting for housing and health schemes.

Attendance and liveness checks now use facial recognition in wage and pension programmes.

Tax and indirect tax administration use analytics to select cases for scrutiny.

The IndiaAI Mission includes a pillar on safe and trusted artificial intelligence.

The accountability questions

A model may never use caste, and still use occupation or locality as a proxy for it.

The Dutch court struck down the SyRI welfare risk system in 2020 on privacy grounds.

Australia's Robodebt scheme raised unlawful debts and led to a royal commission in 2023.

A wrongly excluded person often never learns why, so the error is never appealed.

Proprietary models make the reasoning unavailable even to the department that bought it.

Function creep occurs when data gathered for one scheme is reused for another.

Exclusion is invisible in the data, since the excluded person leaves no record.

Safeguards and standards

NITI Aayog published Responsible Artificial Intelligence principles in 2021.

Puttaswamy, 2018 barred denial of a benefit for authentication failure alone.

Procurement conditions can require explainability and access to the decision logic.

A sunset clause forces a fresh justification before a system is continued.

Piloting with an independent evaluation should precede any state wide rollout.

Grievance redress must sit outside the department that runs the system.

UPI now handles well over a thousand crore transactions a month. Aadhaar authentication runs into billions. Reach is no longer the problem.

Why exclusion remains

A phone is not always owned: Mobile ownership is lower among women and the poorest households. A woman's account may then be operated by someone else.

Documents assume a settled life: A migrant tenant has no local address proof. One Nation One Ration Card exists because entitlements were tied to one place.

The last mile is a person: A banking correspondent short of cash turns a transfer into a journey. Rural withdrawal often needs two attempts.

Remedies

Always keep a manual route: Puttaswamy, 2018 barred denial of a benefit for authentication failure alone.

Make the gap visible: Publish authentication failure rates by district. An excluded person leaves no record, so nobody counts them.

The rails widened reach and narrowed the exit. What remains is small in proportion and severe in effect. It falls on those with no other route.

Decode

Two demands. Causes of residual exclusion, then remedies.

Implicit demand

Residual means what remains after scale. Do not argue against digital delivery itself.

Architecture

Define the term → causes → remedies → verdict.

Articulation

Anchor: Aadhaar, UPI, biometric failure, Puttaswamy, common service centres, Bhashini. Money line: small in proportion and severe in effect.

The system

The three layers are identity, payment and data, often called the India Stack.

UPI, DigiLocker, CoWIN, FASTag and ONDC are built on the same rails.

The Account Aggregator framework moves consent based financial data.

Aadhaar Enabled Payment System allows withdrawal at a banking correspondent.

BharatNet was meant to carry connectivity to every gram panchayat.

Who is left out

Fingerprints of manual labourers and the elderly are often unreadable, which is why face authentication was added to pensions.

Manual labourers, whose fingerprints wear down with the work they do.

Elderly pensioners, for whom liveness checks fail most often.

Migrant workers, who lack local address proof and portable records.

Women, who face a gap in both device ownership and independent account use.

Persons with disabilities, for whom most interfaces are not designed.

Households in areas where the network itself is unreliable.

What has worked

One Nation One Ration Card made entitlements portable across states.

Face authentication was introduced after repeated fingerprint failures in pensions.

Common Service Centres provide assisted access in rural areas.

Bhashini supports translation and voice interfaces in Indian languages.

Publishing failure data is the single change that would make exclusion visible.

Social audit entered MGNREGA in 2005 and the food security law in 2013. Two decades on, many panchayats have never been audited. Proved irregularities rarely end in recovery.

Why it stays weak

The audited control the auditor: Social Audit Units draw staff and budget from the implementing department. Directors have been removed by the very department under audit.

The money does not arrive: Half of one per cent of scheme spending is earmarked for the unit. States frequently withhold it, so audits run on skeleton staff.

Findings lead nowhere: There is no duty to file an action taken report within a fixed period. Villagers who testify see nothing happen, and stop coming.

The auditor is unprotected: Village resource persons face pressure from local contractors and officials. The Whistle Blowers Protection Act, 2014 has never been brought into force.

The gram sabha can be captured: Quorum rules are weak, so a dominant group decides who speaks. Records also arrive on the day, leaving no time to verify.

Measures

Cut the dependence: Give the unit a fixed tenure director and a separate budget line. Release the earmarked share automatically, without departmental approval.

Make a reply compulsory: Require a published action taken report, with recovery ordered where loss is proved. Link further fund release to audit compliance.

Widen and standardise: Meghalaya, 2017 extended social audit to many schemes by law. The Comptroller and Auditor General has issued standards for the exercise.

Social audit fails not because villages cannot audit, but because the audited control the auditor. Independence of the unit, and a duty to reply, would change more than any training.

Decode

Two demands. Why it is weak, then measures to strengthen it.

Implicit demand

Downward accountability is the key phrase. The failure is structural, not technical.

Architecture

Define it → the design → why it stays weak → measures → verdict.

Articulation

Anchor: Section 17 of MGNREGA, Social Audit Units, MKSS, the CAG standards, the Meghalaya law. Money line: the audited control the auditor.

The design

The MGNREGA Audit of Scheme Rules, 2011 govern how the audit is conducted.

Half of one per cent of scheme expenditure is earmarked for the social audit unit.

Section 28 of the National Food Security Act, 2013 also provides for social audit.

The MKSS public hearings at Beawar in the 1990s began the practice in India.

The Comptroller and Auditor General issued auditing standards for social audit in 2016.

Meghalaya passed the first state law covering social audit across many schemes.

Why it stays weak

Directors of audit units have been appointed and removed by the implementing department.

Several states have conducted audits in only a fraction of their panchayats.

Village resource persons are paid little and trained briefly for demanding work.

Documents are often supplied on the day of the hearing, leaving no time to verify.

Gram sabha quorum rules are weak, so a small group can dominate proceedings.

Findings are not linked to any consequence, so the exercise looks futile to villagers.

Measures and models

Andhra Pradesh built one of the earliest and largest social audit units.

Proactive disclosure under Section 4 of the Right to Information Act supports verification.

Machine readable scheme data makes independent checking possible before the hearing.

Linking further fund release to audit compliance creates a real incentive.

Findings can feed into the annual performance appraisal of responsible officers.

The Whistle Blowers Protection Act, 2014 remains largely unoperational.

The 2024 advertisement for lateral posts was withdrawn within days. The objection was that a single post cannot carry reservation.

The case for

Depth a rotation cannot build: Regulating finance or energy needs years in the sector. A short posting cannot produce it.

A vacancy that is real: Central staffing runs short at director level. The posts were created against that gap.

The case against

Equity: A single post cannot be divided, so reservation does not apply. It narrows a route of constitutional weight.

Short horizons and old loyalties: A three year contract ends before the outcome appears. The entrant may also regulate his old industry.

Safeguards if expanded

Select openly, cluster the posts: Selection through the UPSC, with published criteria. Advertise posts in groups, so reservation applies.

Build the same skill inside: Domain cadres and Mission Karmayogi build the same skill inside.

Lateral entry answers a real deficit, but treats a symptom. Without specialisation built within, such posts will always be filled from outside.

Decode

Three demands. Arguments for, arguments against, then safeguards.

Implicit demand

The equity objection is the strongest one. It must be stated, not avoided.

Architecture

What it is → the case for → the case against → safeguards → verdict.

Articulation

Anchor: the 2018 recruitment, reservation in single post advertisements, UPSC selection, Mission Karmayogi. Money line: it answers a real deficit, but treats a symptom.

The record

Lateral recruitment at joint secretary level began with an advertisement in 2018.

Selection has been conducted by the Union Public Service Commission.

Later rounds added director and deputy secretary level posts.

A 2024 advertisement was withdrawn after objections about reservation.

Appointments are usually on contract for three years, extendable to five.

The recommendations behind it

The Second ARC tenth report proposed specialisation in senior appointments.

The Baswan Committee, 2016 examined recruitment to the civil services.

NITI Aayog recommended lateral entry in its action agenda in 2017.

The Surinder Nath Committee and the Hota Committee both examined domain assignment.

Domain assignment for career officers is the internal version of the same idea.

The objections in detail

Single post recruitment escapes reservation, since a post cannot be divided.

Career officers see promotion prospects narrow at the level they are due to reach.

Private sector entrants may lack familiarity with public law and procedure.

A contract officer has weaker protection, and may find it harder to say no.

Conflict of interest rules for entering and leaving service remain thin.

Transparency of the selection process matters more when the pool is small.

SEBI, TRAI and the electricity commissions are all created by statute. Each has a fixed term and stated removal grounds. Independence is lost in the structure, not in the statute.

Appointments

The executive picks its own regulator: The search committee for financial regulators is led by the Cabinet Secretary. There is no parliamentary confirmation and no opposition presence.

A vacancy is itself a lever: The Competition Commission once lacked quorum and could not decide cases for months. Delay in filling posts achieves what removal cannot.

Tenure

Short terms, renewable at discretion: A three year term with a possible second one shapes the final year. An official hoping for renewal does not provoke the ministry.

The budget is annual and the directions bind: Section 25 of the TRAI Act allows binding policy directions. A regulator that returns each year for funds negotiates its autonomy.

Post retirement employment

The next employer is the state or the industry: Tribunals, commissions and advisory posts follow retirement. So do boards in the regulated sector, since most statutes set no cooling off.

The government wears two hats: It writes policy and owns large firms in power, telecom and oil. The regulator then decides against its own appointing authority.

Measures

One term, broadly chosen: Five years, not renewable, selected by a panel including the opposition and a judicial member.

Money and exit, both settled: Fund the regulator from a dedicated levy. Bar employment in a regulated entity for two years.

Independence written in a statute is only a promise. It becomes real when the regulator needs nothing from the government. Not a job, a budget or a second term.

Decode

The stem names three heads. Answer under those three, then add measures.

Implicit demand

Structural means built into the design. Do not blame individuals.

Architecture

Define independence → appointments → tenure → post retirement → the deeper conflict → measures → verdict.

Articulation

Anchor: search committees led by officials, the Competition Commission quorum, policy directions, cooling off. Money line: independence is real when the regulator needs nothing from the government.

Appointments

The Financial Sector Regulatory Appointments Search Committee is headed by the Cabinet Secretary.

Different regulators follow different selection routes, with no common standard.

Vacancies at the Competition Commission and at tribunals have blocked decisions.

The Financial Sector Legislative Reforms Commission, 2013 proposed a common governance framework.

Comparative note: some countries require legislative confirmation of a regulator.

Tenure and control

Section 25 of the TRAI Act allows the government to issue directions on policy questions.

The Electricity Act, 2003 contains a similar provision for directions to commissions.

Salaries and conditions of service are usually fixed by the government.

Appellate tribunals have depended on the ministry for staff and accommodation.

Rojer Mathew, 2019 and the Madras Bar Association cases dealt with tribunal independence.

A regulator that must return to the ministry for funds is negotiating every year.

Post retirement and reform

There is no uniform cooling off period across Indian regulators.

Movement to a regulated firm after office is the classic revolving door problem.

The government is often the owner of large firms in the sectors it regulates.

Separating ownership from policy is the structural fix, discussed for power and telecom.

Regulatory impact assessment before a rule would make reasoning public.

A standing parliamentary committee review would supply accountability without control.